Knowledge Partitioning in Outsourced Software Development: A Field Study

The outsourced software development process has traditionally relied on a requirements-driven black-box approach for transferring knowledge of customer needs to vendors. When this approach is feasible, the need for the customer and the vendor to deeply understand each others’ knowledge domain is limited. We describe this as symmetric division of knowledge. However, asymmetric overlaps in knowledge are necessary at the vendor-customer boundary in projects involving conceptual or process newness. In this study, we examine the conditions under which overlaps in knowledge at the vendor-customer boundary are necessary for enhancing the development process in outsourcing relationships. We develop and test a model using data collected in a large-scale field study of 209 software projects in 209 software development organizations belonging to three of the largest global software consortia. The study makes three contributions: (1) we empirically demonstrate that it is more important for a vendor to possess a higher level of business knowledge in conceptually new projects and for the customer to have a higher level of technical knowledge when the project involves process newness, (2) we assess the effectiveness of various integrating mechanisms, and (3) we show that there are potential downsides to blindly increasing vendor-customer overlaps in knowledge beyond those that have traditionally characterized software development.